Satti Saghar

Artificial Intelligence graduate seeking a Junior AI/ML Engineer role. Hands-on experience designing and deploying machine learning, deep learning, and computer vision systems. Proficient in Python, TensorFlow, Scikit-Learn, Flask REST APIs, EEG biosignal processing, and multimodal AI architectures.

Built a production-grade end-to-end AI solution integrating facial expression analysis, voice signal classification, and EEG-based deception detection for real-time inference. Strong foundation in model training, feature engineering, data preprocessing, and full-stack AI development.

Final Year Project - BIIT, PMAS Arid Agriculture University - Islamabad, Pakistan

(2025-01 - 2026-05)

• Engineered an end-to-end multimodal AI pipeline fusing EEG biosignals, facial expression analysis, and voice streams for real-time deception detection, integrating 3 independent machine learning models across 3 modalities.
• Trained and optimized CNN, LSTM, and XGBoost classifiers and designed an ensemble voting mechanism to fuse predictions across modalities, achieving improved classification accuracy over single-model baselines.
• Extracted 5 EEG frequency-band features (Alpha, Beta, Gamma, Theta, Delta) and MFCC audio features using Librosa, applying advanced feature engineering techniques for signal-level classification.
• Deployed DeepFace and OpenCV pipeline for real-time facial landmark detection, processing live video frames for expression-based feature extraction integrated into the inference layer.
• Built Flask REST API backend connected to SQL Server and developed a React Native mobile frontend with synchronized triple-stream recording via FFmpeg, reducing manual data collection effort by over 60%.

Computer Vision Project - Personal Project

(2024-01 - 2024-12)

• Built a CNN-based object detection and image classification model to identify trucks in live video streams under varied lighting, occlusion, and real-world traffic conditions with high precision.
• Applied OpenCV preprocessing pipelines including frame extraction, normalization, and data augmentation across 3+ traffic scenario types, improving model generalization and classification performance across diverse datasets.

Bachelor of Science - Artificial Intelligence - BIIT, PMAS Arid Agriculture University (2026-05)